I have argued over and over (for example, here, here, and here) that large language models (LLMs) create an illusion of intelligence but are, in fact, prone to making logical and factual errors. That makes it risky to rely on them when the costs of mistakes are substantial. As a lifelong finance professor and investor, I have noted that LLMs can give bad financial advice and that AI-powered ETFs and mutual funds have substantially underperformed the S&P 500.
Several correspondents have pointed out that humans make mistakes too, which is undeniable. For example, the Wall Street Journal recently published an op-ed with the provocative title, “The Case Against 30-Year Mortgages.” The article itself was even more provocative — and wrongheaded. The author argued that the reason millennials are struggling to buy homes is “an insidious financial instrument so predatory and deceptive that it has warped the housing market for nearly a century. Ladies and gentlemen, I present the 30-year mortgage.” The author went on: “The mortgage isn’t the foundation of the American Dream. It’s the scam of the century: a loan so exploitative it required a federal law to disguise its true nature.”
Understanding mortgages
What is the mortgage’s true nature? When homebuyers obtain a mortgage, they have to pay interest. With a 30-year mortgage, the total interest payments are substantial. For a specific example, the WSJ author considers a $400,000 house with a 20% down payment and a 30-year mortgage at a 6% loan rate. The total payments over 30 years amount to $690,000; so the buyer “effectively pays for the home almost twice.” Shocking! You have to pay more interest the more you borrow and the longer you borrow. Carried to its logical extreme, the author would minimize interest payments by never borrowing. In reality, debt can be a good four-letter word.
The author bemoans the focus on a loan’s annual percentage rate (APR):
It’s duplicitous to explain a loan in terms of APR. While a 5% APR may sound benign — it’s lower than the APR for a car loan or a credit card — the comparison is meaningless because it doesn’t account for time. A car loan lasts five years, and a credit card can be paid off at will, meaning the interest has little time to compound.
Really? A 5% mortgage is worse than a 12% car loan or 20% credit card rate?
The APR is, in fact, critical for deciding whether debt is a good or bad four-letter word. It is tempting to put borrowing and investment decisions in separate mental buckets, but they are intimately related. Borrowing at 5% to invest at 10% is profitable and is more profitable the more you borrow and the longer you borrow. This means that a household that is earning a 10% return on its investments, offered a 2% mortgage, should make a low down payment and choose a long-term mortgage — exactly the opposite of the WSJ recommendation. The homeowners should pay off their car loans and credit card debts and celebrate their low-interest mortgage.
What are a home buyer’s alternatives to a 30-year mortgage?
One alternative is the author’s suggestion that buyers pay cash. When I wrote about housing prices in the mid-2000s, a gentleman in Connecticut told me that everyone he knows pays cash for their homes because it is a sign of weakness to borrow money. No, I’m not kidding. Today, a few Silicon Valley millennials may be able to pay cash but the vast majority of millennials can’t.
A second alternative is a shorter-term mortgage. The shorter the loan, the less interest is paid and that is the metric the author has his eyes fixed on. The problem is that, in the author’s $400,000 home example, the monthly payments increase from $1,919 with a 30-year mortgage to $3,553 with a 10-year mortgage, and $6,187 with a 5-year mortgage. Using the rule that a household should not spend more than 28% of its gross income on mortgage payments, taxes, and insurance, and assuming $4,000 in annual property taxes and $2,000 for insurance, a home buyer would need an annual income of $104,000 to qualify for a 30-year mortgage. Reducing the length of the mortgage increases the required annual income to $173,700 with a 10-year mortgage and $286,586 with a 5-year mortgage — effectively precluding home ownership for most millennials.
A third alternative is to rent instead of buy. Which brings us to the real question: We have to live somewhere and we should consider whether it is it cheaper to buy a home or to rent one. The three most important words in real estate are, of course, location, location, location. So the decision depends on the details.
Looking at an example…
In 2005, in the midst of the so-called housing bubble, Margaret Smith and I looked at ten US cities and found some were bubbly and others (like Indianapolis) were not. One of our Indianapolis homes was a 3-bedroom, 3-bathroom, 1,912-square-foot home in Fishers, Indiana, a suburb of Indianapolis. Money magazine has ranked Fishers as among the top 50 places to live in the United States multiple times. In 2017, Fishers was rated #1 in the country; in 2019, it was rated #3.
This home was purchased for $135,000 on April 27, 2005, and rented for $1,250/month on June 1, 2005. The table shows the first-year net cash flow for someone who bought the house and lived in it instead of renting it, either paying cash or using a 30-year mortgage with a 5.7% APR:
No Mortgage Mortgage
Rent savings 15,000 15,000
Mortgage payment 0 –7,522
Property tax –2619 –2619
Tax savings 733 2,447
Insurance –334 –334
Maintenance –1,350 –1,350
Net cash flow $11,430 $5,622
The $11,430 first-year net cash flow for the all-cash buyer represents an 8.5% return on the $135,000 purchase price. Instead of paying cash, taking a mortgage — borrowing at 5.7% to invest at 8.5% — increases the buyer’s rate of return from 8.5% to an astonishing 20.8% (a $5,622 return on the $27,000 down payment). The cash flow increases every year as the rent savings increase while the mortgage payments are fixed and then stop after 30 years. There was no housing bubble in Fishers!
This home has not been resold but Zillow estimates its current market value to be $334,000. If this house were sold today for $334,000 with, say, $15,000 in closing costs, the seller would net $262,064 after paying off the mortgage balance, giving a 12.0% annual return on the initial down payment. Add in the annual net cash flow from owning instead of renting and the annual rate of return jumps to 29.9%.
The total mortgage payments over 30 years are $225,659, which is more than double the $108,000 that was borrowed. The WSJ author would scream “Scam!” Yet, this was far from a scam. The homebuyer with a 30-year mortgage has made a 29.9% annual return over 20 years and would have made an even higher return if more money had been borrowed or if the mortgage had been even longer term.
What can we take away from this?
Not every home is so profitable. It depends on the mortgage rate, rent savings, and so on. The takeaway is that, when considering a mortgage (or any loan), the total payments are the wrong thing to consider. The correct comparison is the return on the borrowed money versus the loan’s APR. If the return on the borrowed money is higher than the APR, larger and longer loans are financially advantageous — which is why the most profitable investment many people will ever make is buying a home with a long-term mortgage.
Note: How did ChatGPT do on this question? I asked GPT-5, “I need to borrow $300,000 to buy a home. Should I take out a 10-year or 30-year mortgage? Both loans have a 2.0% APR with no points.” Despite the irresistibly low APR, GPT-5 made the same total-payments error as the WSJ author, recommending the 10-year mortgage as “the financially superior option” because it “saves you about $68,000 in interest versus the 30-year.”
